Résumé
The great Tohoku-Oki mega-thrust earthquake, which was significantly underestimated by seismologists, gives us an opportunity to understand the complex subduction process and associated seismic hazard. Our Japan-France (ANR-JST) research cooperative 'PALET' project of Tohoku-oki paleoseismology, addresses the issue of long-term faulting behaviour and earthquake prevision of the NE Japan coastline. The aim is to reveal the seismic cycle of M9 events, by investigating 1) paleo-tsunami record in Holocene deposits, 2) long-term behaviour of directly triggered inland normal faults, and 3) long-term deformation using marine terraces and comparison with the 2011 coseismic movements. Following the pioneering papers on the 869 A.D. Jogan tsunami in the Sendai to Fukushima region, we focus here on the Sanriku coast, north of the 2011 source. Previous paleotsunami studies (Haraguchi & Ishibe, 2009) performed at six sites are added to new surveys from Kesennuma area. Seven well-preserved paleo-tsunami horizons from 1,000 and 6,000 y.B.P were found at six onshore sites, yielding 500-700 years of recurrence intervals of extremely large tsunamis. A 35-m coring sample recovered from the Ozuchi Bay instead preserved 22 units of tsunami deposits during the past 8,000 years suggesting more frequent but relatively smaller tsunami events. Seismic hazard onshore influenced by the Tohoku-oki earthquake has been dramatically raised. A typical case is the Mw 6.6 Iwaki earthquake of 11 April 2011 with surface faulting. Our stress calculations suggest that the Tohoku-Oki mainshock flipped stress field in the fore-arc region and promoted the normal faulting activity that had been extremely dormant before 2011. We mapped two sub-parallel surface ruptures (15-km long each) and then conducted paleoseismic trenches. We found evidence for the penultimate earthquake that occurred sometime between 12,600 and 17,400 y.B.P. However, since several other normal faults in the region share the similar active geomorphic features, ideal paleoseimic records from all the faults could indirectly reveal the history of offshore M9 events. To evaluate post-Tohoku hazard, we perform detail comparisons between the 2011 vertical coastal movements and the altitudes of marine terraces of MIS stages 5e ( approximately 120 ka) and MIS 1 ( approximately 6 ka) along the entire pacific coast. Long-term uplift rate of 0.3-0.5 mm/yr in Aomori and Fukushima regions, and 0.1-0.2 mm/yr in Iwate region apparently contradict to the 2011 subsidence of up to 140 cm, for which we may have to consider the combined effects of the interseismic deformation and/or adjacent mega-thrust events. However, the Sanriku coast, characterized as a ria coastline, is in still controversial about its long-term behavior, which so far leads to three contending scenarios, 1) the 2011 event was typical, contributing long-term subsidence, 2) a significant interseismic contribution to long term coastal uplift, or 3) a hidden huge event may occur to uplift the coastal region.